spellcheck not displaying potential spelling errors

[Rob557]

Sometimes when using Calibre's Edit Book's spell checker I encounter ePubs that display no potential spelling errors even when I can see there are some errors, which contrasts with most epubs where the spell checker will show hundreds of potential spelling errors even when a review of the list confirms the words are not spelled incorrectly but just a bit more 'sophisticated'.

I haven't figured out what is unique about those problematic ePubs that display no potential spelling errors (edit check indicates 'no problems' for the ePub structure itself), so I have set up a small sample extract ePub and for good measure have prefaced the extract with a few words that should definitely appear in the spell checker display.

The sample "test author - test book.epub" is at the following link:
http://www73.zippyshare.com/v/XKrw3170/file.html

Thanks in advance for any suggestions on what I would need to change within the ePub so that the Calibre spellchecker will display the potential spelling errors.

[theducks]

Originally Posted by Rob557

Sometimes when using Calibre's Edit Book's spell checker I encounter ePubs that display no potential spelling errors even when I can see there are some errors, which contrasts with most epubs where the spell checker will show hundreds of potential spelling errors even when a review of the list confirms the words are not spelled incorrectly but just a bit more 'sophisticated'.

I haven't figured out what is unique about those problematic ePubs that display no potential spelling errors (edit check indicates 'no problems' for the ePub structure itself), so I have set up a small sample extract ePub and for good measure have prefaced the extract with a few words that should definitely appear in the spell checker display.

The sample "test author - test book.epub" is at the following link:
http://www73.zippyshare.com/v/XKrw3170/file.html

Thanks in advance for any suggestions on what I would need to change within the ePub so that the Calibre spellchecker will display the potential spelling errors.

The book claims the Language is RUSSIAN

BTW Sigil seems to ignore the Language and does find errors

[Rob557]

Originally Posted by theducks

The book claims the Language is [B]RUSSIAN[/B

The text is clearly English and the only language indicator I had found in the html code was at the start of titlepage.xhmtl where it says xml:lang="en".

But after reading your comment I looked around some more and now see that if I just look at the basic metadata there is an editable field for language. I've never had cause to look at that before (the language always seemed self-evident ).

I changed that metadata indicator from Russian to English and the spell checker now displays the potential spelling errors okay.

Thank you theducks for solving that problem !!

[theducks]

Originally Posted by Rob557

The text is clearly English and the only language indicator I had found in the html code was at the start of titlepage.xhmtl where it says xml:lang="en".

But after reading your comment I looked around some more and now see that if I just look at the basic metadata there is an editable field for language. I've never had cause to look at that before (the language always seemed self-evident ).

I changed that metadata indicator from Russian to English and the spell checker now displays the potential spelling errors okay.

Thank you theducks for solving that problem !!

FWIW the spell checker Had a Language column
Not only is there English, those outside the USA have their own spellings and still call it english.

I am forever running into French words flagged wrong.
Lots of words with French roots appear in restaurant scenes, assignations...

[Rob557]

Originally Posted by theducks

FWIW the spell checker had a Language column

Yes, agreed, but unfortunately when there was no displayed listing of potential spelling errors, there was also no content in the spell checker display that identified the incorrectly assumed language.

In any case, it is enough to know what the problem is likely to be the next time the spell checker display is empty. Although you raise a good point that maybe the assumed language should be automatically displayed, to help others who might encounter the problem in the future.

FWIW, I recall one time that there were two different languages specified in the spell checker's language column (although any one word showed only one of the languages), some sort of variation of English. I'm not sure how that would have happened because the metadata field allows for only a single language type to be specified, but maybe within each html file within the ePub it is possible to somehow override the metadata-identified language. Anyway, I digress ...

Thanks again.

[amazon author]

Microsoft word catches much more error.

[kovidgoyal]

The spell checker displays both the misspelled word and the language against which that word was checked. In HTML you can specify different languages for different parts of the text using the lang attribute. The overall book language, which is what you set with metadata is simply the default language used when the HTML does not specify a language. So if your book or the HTML in it specifies a language for which you dont have a spell check dictionary installed, you will get no spelling errors.

[Rob557]

Originally Posted by kovidgoyal

In HTML you can specify different languages for different parts of the text using the lang attribute. The overall book language, which is what you set with metadata is simply the default language

Thank you for that clarification Kovid. Makes sense and I see now that replacing <html xmlns="http://www.w3.org/1999/xhtml"> with <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en"> at the start of each html file (and not just appearing in the initial titlepage file) would also have corrected the problem, but less conveniently.

I guess whenever an incorrect default language is specified in the metadata, if the correct language is not specified at the start of EVERY html files then the spellchecker default would kick in sometimes and would not work properly for some sections of the text.

Accordingly, it might be useful in some future modification for the spellchecker to also indicate the default language in the metadata, particularly where the default language has the effect that the spellchecker does not work for some sections of the text. If the spellchecker was literally checking for potential Russian spelling errors then almost every word would display as a potential spelling error, so the spellchecker must have some sort of override to stop that from happening.

Originally Posted by kovidgoyal

... if your book or the HTML in it specifies a language for which you dont have a spell check dictionary installed, you will get no spelling errors.

I guess that would be the override effect for the incorrect default language in this case.